linux/include/linux/sched.h
<<
>>
Prefs
   1#ifndef _LINUX_SCHED_H
   2#define _LINUX_SCHED_H
   3
   4/*
   5 * cloning flags:
   6 */
   7#define CSIGNAL         0x000000ff      /* signal mask to be sent at exit */
   8#define CLONE_VM        0x00000100      /* set if VM shared between processes */
   9#define CLONE_FS        0x00000200      /* set if fs info shared between processes */
  10#define CLONE_FILES     0x00000400      /* set if open files shared between processes */
  11#define CLONE_SIGHAND   0x00000800      /* set if signal handlers and blocked signals shared */
  12#define CLONE_PTRACE    0x00002000      /* set if we want to let tracing continue on the child too */
  13#define CLONE_VFORK     0x00004000      /* set if the parent wants the child to wake it up on mm_release */
  14#define CLONE_PARENT    0x00008000      /* set if we want to have the same parent as the cloner */
  15#define CLONE_THREAD    0x00010000      /* Same thread group? */
  16#define CLONE_NEWNS     0x00020000      /* New namespace group? */
  17#define CLONE_SYSVSEM   0x00040000      /* share system V SEM_UNDO semantics */
  18#define CLONE_SETTLS    0x00080000      /* create a new TLS for the child */
  19#define CLONE_PARENT_SETTID     0x00100000      /* set the TID in the parent */
  20#define CLONE_CHILD_CLEARTID    0x00200000      /* clear the TID in the child */
  21#define CLONE_DETACHED          0x00400000      /* Unused, ignored */
  22#define CLONE_UNTRACED          0x00800000      /* set if the tracing process can't force CLONE_PTRACE on this clone */
  23#define CLONE_CHILD_SETTID      0x01000000      /* set the TID in the child */
  24/* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
  25   and is now available for re-use. */
  26#define CLONE_NEWUTS            0x04000000      /* New utsname group? */
  27#define CLONE_NEWIPC            0x08000000      /* New ipcs */
  28#define CLONE_NEWUSER           0x10000000      /* New user namespace */
  29#define CLONE_NEWPID            0x20000000      /* New pid namespace */
  30#define CLONE_NEWNET            0x40000000      /* New network namespace */
  31#define CLONE_IO                0x80000000      /* Clone io context */
  32
  33/*
  34 * Scheduling policies
  35 */
  36#define SCHED_NORMAL            0
  37#define SCHED_FIFO              1
  38#define SCHED_RR                2
  39#define SCHED_BATCH             3
  40/* SCHED_ISO: reserved but not implemented yet */
  41#define SCHED_IDLE              5
  42/* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
  43#define SCHED_RESET_ON_FORK     0x40000000
  44
  45#ifdef __KERNEL__
  46
  47struct sched_param {
  48        int sched_priority;
  49};
  50
  51#include <asm/param.h>  /* for HZ */
  52
  53#include <linux/capability.h>
  54#include <linux/threads.h>
  55#include <linux/kernel.h>
  56#include <linux/types.h>
  57#include <linux/timex.h>
  58#include <linux/jiffies.h>
  59#include <linux/rbtree.h>
  60#include <linux/thread_info.h>
  61#include <linux/cpumask.h>
  62#include <linux/errno.h>
  63#include <linux/nodemask.h>
  64#include <linux/mm_types.h>
  65
  66#include <asm/page.h>
  67#include <asm/ptrace.h>
  68#include <asm/cputime.h>
  69
  70#include <linux/smp.h>
  71#include <linux/sem.h>
  72#include <linux/signal.h>
  73#include <linux/compiler.h>
  74#include <linux/completion.h>
  75#include <linux/pid.h>
  76#include <linux/percpu.h>
  77#include <linux/topology.h>
  78#include <linux/proportions.h>
  79#include <linux/seccomp.h>
  80#include <linux/rcupdate.h>
  81#include <linux/rculist.h>
  82#include <linux/rtmutex.h>
  83
  84#include <linux/time.h>
  85#include <linux/param.h>
  86#include <linux/resource.h>
  87#include <linux/timer.h>
  88#include <linux/hrtimer.h>
  89#include <linux/task_io_accounting.h>
  90#include <linux/latencytop.h>
  91#include <linux/cred.h>
  92#include <linux/llist.h>
  93#include <linux/uidgid.h>
  94
  95#include <asm/processor.h>
  96
  97struct exec_domain;
  98struct futex_pi_state;
  99struct robust_list_head;
 100struct bio_list;
 101struct fs_struct;
 102struct perf_event_context;
 103struct blk_plug;
 104
 105/*
 106 * List of flags we want to share for kernel threads,
 107 * if only because they are not used by them anyway.
 108 */
 109#define CLONE_KERNEL    (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
 110
 111/*
 112 * These are the constant used to fake the fixed-point load-average
 113 * counting. Some notes:
 114 *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
 115 *    a load-average precision of 10 bits integer + 11 bits fractional
 116 *  - if you want to count load-averages more often, you need more
 117 *    precision, or rounding will get you. With 2-second counting freq,
 118 *    the EXP_n values would be 1981, 2034 and 2043 if still using only
 119 *    11 bit fractions.
 120 */
 121extern unsigned long avenrun[];         /* Load averages */
 122extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
 123
 124#define FSHIFT          11              /* nr of bits of precision */
 125#define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
 126#define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
 127#define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
 128#define EXP_5           2014            /* 1/exp(5sec/5min) */
 129#define EXP_15          2037            /* 1/exp(5sec/15min) */
 130
 131#define CALC_LOAD(load,exp,n) \
 132        load *= exp; \
 133        load += n*(FIXED_1-exp); \
 134        load >>= FSHIFT;
 135
 136extern unsigned long total_forks;
 137extern int nr_threads;
 138DECLARE_PER_CPU(unsigned long, process_counts);
 139extern int nr_processes(void);
 140extern unsigned long nr_running(void);
 141extern unsigned long nr_uninterruptible(void);
 142extern unsigned long nr_iowait(void);
 143extern unsigned long nr_iowait_cpu(int cpu);
 144extern unsigned long this_cpu_load(void);
 145
 146
 147extern void calc_global_load(unsigned long ticks);
 148extern void update_cpu_load_nohz(void);
 149
 150extern unsigned long get_parent_ip(unsigned long addr);
 151
 152struct seq_file;
 153struct cfs_rq;
 154struct task_group;
 155#ifdef CONFIG_SCHED_DEBUG
 156extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
 157extern void proc_sched_set_task(struct task_struct *p);
 158extern void
 159print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
 160#else
 161static inline void
 162proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 163{
 164}
 165static inline void proc_sched_set_task(struct task_struct *p)
 166{
 167}
 168static inline void
 169print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
 170{
 171}
 172#endif
 173
 174/*
 175 * Task state bitmask. NOTE! These bits are also
 176 * encoded in fs/proc/array.c: get_task_state().
 177 *
 178 * We have two separate sets of flags: task->state
 179 * is about runnability, while task->exit_state are
 180 * about the task exiting. Confusing, but this way
 181 * modifying one set can't modify the other one by
 182 * mistake.
 183 */
 184#define TASK_RUNNING            0
 185#define TASK_INTERRUPTIBLE      1
 186#define TASK_UNINTERRUPTIBLE    2
 187#define __TASK_STOPPED          4
 188#define __TASK_TRACED           8
 189/* in tsk->exit_state */
 190#define EXIT_ZOMBIE             16
 191#define EXIT_DEAD               32
 192/* in tsk->state again */
 193#define TASK_DEAD               64
 194#define TASK_WAKEKILL           128
 195#define TASK_WAKING             256
 196#define TASK_STATE_MAX          512
 197
 198#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
 199
 200extern char ___assert_task_state[1 - 2*!!(
 201                sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
 202
 203/* Convenience macros for the sake of set_task_state */
 204#define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
 205#define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
 206#define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
 207
 208/* Convenience macros for the sake of wake_up */
 209#define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
 210#define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
 211
 212/* get_task_state() */
 213#define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
 214                                 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
 215                                 __TASK_TRACED)
 216
 217#define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
 218#define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
 219#define task_is_dead(task)      ((task)->exit_state != 0)
 220#define task_is_stopped_or_traced(task) \
 221                        ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
 222#define task_contributes_to_load(task)  \
 223                                ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
 224                                 (task->flags & PF_FROZEN) == 0)
 225
 226#define __set_task_state(tsk, state_value)              \
 227        do { (tsk)->state = (state_value); } while (0)
 228#define set_task_state(tsk, state_value)                \
 229        set_mb((tsk)->state, (state_value))
 230
 231/*
 232 * set_current_state() includes a barrier so that the write of current->state
 233 * is correctly serialised wrt the caller's subsequent test of whether to
 234 * actually sleep:
 235 *
 236 *      set_current_state(TASK_UNINTERRUPTIBLE);
 237 *      if (do_i_need_to_sleep())
 238 *              schedule();
 239 *
 240 * If the caller does not need such serialisation then use __set_current_state()
 241 */
 242#define __set_current_state(state_value)                        \
 243        do { current->state = (state_value); } while (0)
 244#define set_current_state(state_value)          \
 245        set_mb(current->state, (state_value))
 246
 247/* Task command name length */
 248#define TASK_COMM_LEN 16
 249
 250#include <linux/spinlock.h>
 251
 252/*
 253 * This serializes "schedule()" and also protects
 254 * the run-queue from deletions/modifications (but
 255 * _adding_ to the beginning of the run-queue has
 256 * a separate lock).
 257 */
 258extern rwlock_t tasklist_lock;
 259extern spinlock_t mmlist_lock;
 260
 261struct task_struct;
 262
 263#ifdef CONFIG_PROVE_RCU
 264extern int lockdep_tasklist_lock_is_held(void);
 265#endif /* #ifdef CONFIG_PROVE_RCU */
 266
 267extern void sched_init(void);
 268extern void sched_init_smp(void);
 269extern asmlinkage void schedule_tail(struct task_struct *prev);
 270extern void init_idle(struct task_struct *idle, int cpu);
 271extern void init_idle_bootup_task(struct task_struct *idle);
 272
 273extern int runqueue_is_locked(int cpu);
 274
 275#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
 276extern void select_nohz_load_balancer(int stop_tick);
 277extern void set_cpu_sd_state_idle(void);
 278extern int get_nohz_timer_target(void);
 279#else
 280static inline void select_nohz_load_balancer(int stop_tick) { }
 281static inline void set_cpu_sd_state_idle(void) { }
 282#endif
 283
 284/*
 285 * Only dump TASK_* tasks. (0 for all tasks)
 286 */
 287extern void show_state_filter(unsigned long state_filter);
 288
 289static inline void show_state(void)
 290{
 291        show_state_filter(0);
 292}
 293
 294extern void show_regs(struct pt_regs *);
 295
 296/*
 297 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
 298 * task), SP is the stack pointer of the first frame that should be shown in the back
 299 * trace (or NULL if the entire call-chain of the task should be shown).
 300 */
 301extern void show_stack(struct task_struct *task, unsigned long *sp);
 302
 303void io_schedule(void);
 304long io_schedule_timeout(long timeout);
 305
 306extern void cpu_init (void);
 307extern void trap_init(void);
 308extern void update_process_times(int user);
 309extern void scheduler_tick(void);
 310
 311extern void sched_show_task(struct task_struct *p);
 312
 313#ifdef CONFIG_LOCKUP_DETECTOR
 314extern void touch_softlockup_watchdog(void);
 315extern void touch_softlockup_watchdog_sync(void);
 316extern void touch_all_softlockup_watchdogs(void);
 317extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
 318                                  void __user *buffer,
 319                                  size_t *lenp, loff_t *ppos);
 320extern unsigned int  softlockup_panic;
 321void lockup_detector_init(void);
 322#else
 323static inline void touch_softlockup_watchdog(void)
 324{
 325}
 326static inline void touch_softlockup_watchdog_sync(void)
 327{
 328}
 329static inline void touch_all_softlockup_watchdogs(void)
 330{
 331}
 332static inline void lockup_detector_init(void)
 333{
 334}
 335#endif
 336
 337#ifdef CONFIG_DETECT_HUNG_TASK
 338extern unsigned int  sysctl_hung_task_panic;
 339extern unsigned long sysctl_hung_task_check_count;
 340extern unsigned long sysctl_hung_task_timeout_secs;
 341extern unsigned long sysctl_hung_task_warnings;
 342extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
 343                                         void __user *buffer,
 344                                         size_t *lenp, loff_t *ppos);
 345#else
 346/* Avoid need for ifdefs elsewhere in the code */
 347enum { sysctl_hung_task_timeout_secs = 0 };
 348#endif
 349
 350/* Attach to any functions which should be ignored in wchan output. */
 351#define __sched         __attribute__((__section__(".sched.text")))
 352
 353/* Linker adds these: start and end of __sched functions */
 354extern char __sched_text_start[], __sched_text_end[];
 355
 356/* Is this address in the __sched functions? */
 357extern int in_sched_functions(unsigned long addr);
 358
 359#define MAX_SCHEDULE_TIMEOUT    LONG_MAX
 360extern signed long schedule_timeout(signed long timeout);
 361extern signed long schedule_timeout_interruptible(signed long timeout);
 362extern signed long schedule_timeout_killable(signed long timeout);
 363extern signed long schedule_timeout_uninterruptible(signed long timeout);
 364asmlinkage void schedule(void);
 365extern void schedule_preempt_disabled(void);
 366extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
 367
 368struct nsproxy;
 369struct user_namespace;
 370
 371/*
 372 * Default maximum number of active map areas, this limits the number of vmas
 373 * per mm struct. Users can overwrite this number by sysctl but there is a
 374 * problem.
 375 *
 376 * When a program's coredump is generated as ELF format, a section is created
 377 * per a vma. In ELF, the number of sections is represented in unsigned short.
 378 * This means the number of sections should be smaller than 65535 at coredump.
 379 * Because the kernel adds some informative sections to a image of program at
 380 * generating coredump, we need some margin. The number of extra sections is
 381 * 1-3 now and depends on arch. We use "5" as safe margin, here.
 382 */
 383#define MAPCOUNT_ELF_CORE_MARGIN        (5)
 384#define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
 385
 386extern int sysctl_max_map_count;
 387
 388#include <linux/aio.h>
 389
 390#ifdef CONFIG_MMU
 391extern void arch_pick_mmap_layout(struct mm_struct *mm);
 392extern unsigned long
 393arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
 394                       unsigned long, unsigned long);
 395extern unsigned long
 396arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
 397                          unsigned long len, unsigned long pgoff,
 398                          unsigned long flags);
 399extern void arch_unmap_area(struct mm_struct *, unsigned long);
 400extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
 401#else
 402static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
 403#endif
 404
 405
 406extern void set_dumpable(struct mm_struct *mm, int value);
 407extern int get_dumpable(struct mm_struct *mm);
 408
 409/* mm flags */
 410/* dumpable bits */
 411#define MMF_DUMPABLE      0  /* core dump is permitted */
 412#define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
 413
 414#define MMF_DUMPABLE_BITS 2
 415#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
 416
 417/* coredump filter bits */
 418#define MMF_DUMP_ANON_PRIVATE   2
 419#define MMF_DUMP_ANON_SHARED    3
 420#define MMF_DUMP_MAPPED_PRIVATE 4
 421#define MMF_DUMP_MAPPED_SHARED  5
 422#define MMF_DUMP_ELF_HEADERS    6
 423#define MMF_DUMP_HUGETLB_PRIVATE 7
 424#define MMF_DUMP_HUGETLB_SHARED  8
 425
 426#define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
 427#define MMF_DUMP_FILTER_BITS    7
 428#define MMF_DUMP_FILTER_MASK \
 429        (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
 430#define MMF_DUMP_FILTER_DEFAULT \
 431        ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
 432         (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
 433
 434#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
 435# define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
 436#else
 437# define MMF_DUMP_MASK_DEFAULT_ELF      0
 438#endif
 439                                        /* leave room for more dump flags */
 440#define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
 441#define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
 442#define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
 443
 444#define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
 445
 446struct sighand_struct {
 447        atomic_t                count;
 448        struct k_sigaction      action[_NSIG];
 449        spinlock_t              siglock;
 450        wait_queue_head_t       signalfd_wqh;
 451};
 452
 453struct pacct_struct {
 454        int                     ac_flag;
 455        long                    ac_exitcode;
 456        unsigned long           ac_mem;
 457        cputime_t               ac_utime, ac_stime;
 458        unsigned long           ac_minflt, ac_majflt;
 459};
 460
 461struct cpu_itimer {
 462        cputime_t expires;
 463        cputime_t incr;
 464        u32 error;
 465        u32 incr_error;
 466};
 467
 468/**
 469 * struct task_cputime - collected CPU time counts
 470 * @utime:              time spent in user mode, in &cputime_t units
 471 * @stime:              time spent in kernel mode, in &cputime_t units
 472 * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
 473 *
 474 * This structure groups together three kinds of CPU time that are
 475 * tracked for threads and thread groups.  Most things considering
 476 * CPU time want to group these counts together and treat all three
 477 * of them in parallel.
 478 */
 479struct task_cputime {
 480        cputime_t utime;
 481        cputime_t stime;
 482        unsigned long long sum_exec_runtime;
 483};
 484/* Alternate field names when used to cache expirations. */
 485#define prof_exp        stime
 486#define virt_exp        utime
 487#define sched_exp       sum_exec_runtime
 488
 489#define INIT_CPUTIME    \
 490        (struct task_cputime) {                                 \
 491                .utime = 0,                                     \
 492                .stime = 0,                                     \
 493                .sum_exec_runtime = 0,                          \
 494        }
 495
 496/*
 497 * Disable preemption until the scheduler is running.
 498 * Reset by start_kernel()->sched_init()->init_idle().
 499 *
 500 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
 501 * before the scheduler is active -- see should_resched().
 502 */
 503#define INIT_PREEMPT_COUNT      (1 + PREEMPT_ACTIVE)
 504
 505/**
 506 * struct thread_group_cputimer - thread group interval timer counts
 507 * @cputime:            thread group interval timers.
 508 * @running:            non-zero when there are timers running and
 509 *                      @cputime receives updates.
 510 * @lock:               lock for fields in this struct.
 511 *
 512 * This structure contains the version of task_cputime, above, that is
 513 * used for thread group CPU timer calculations.
 514 */
 515struct thread_group_cputimer {
 516        struct task_cputime cputime;
 517        int running;
 518        raw_spinlock_t lock;
 519};
 520
 521#include <linux/rwsem.h>
 522struct autogroup;
 523
 524/*
 525 * NOTE! "signal_struct" does not have its own
 526 * locking, because a shared signal_struct always
 527 * implies a shared sighand_struct, so locking
 528 * sighand_struct is always a proper superset of
 529 * the locking of signal_struct.
 530 */
 531struct signal_struct {
 532        atomic_t                sigcnt;
 533        atomic_t                live;
 534        int                     nr_threads;
 535
 536        wait_queue_head_t       wait_chldexit;  /* for wait4() */
 537
 538        /* current thread group signal load-balancing target: */
 539        struct task_struct      *curr_target;
 540
 541        /* shared signal handling: */
 542        struct sigpending       shared_pending;
 543
 544        /* thread group exit support */
 545        int                     group_exit_code;
 546        /* overloaded:
 547         * - notify group_exit_task when ->count is equal to notify_count
 548         * - everyone except group_exit_task is stopped during signal delivery
 549         *   of fatal signals, group_exit_task processes the signal.
 550         */
 551        int                     notify_count;
 552        struct task_struct      *group_exit_task;
 553
 554        /* thread group stop support, overloads group_exit_code too */
 555        int                     group_stop_count;
 556        unsigned int            flags; /* see SIGNAL_* flags below */
 557
 558        /*
 559         * PR_SET_CHILD_SUBREAPER marks a process, like a service
 560         * manager, to re-parent orphan (double-forking) child processes
 561         * to this process instead of 'init'. The service manager is
 562         * able to receive SIGCHLD signals and is able to investigate
 563         * the process until it calls wait(). All children of this
 564         * process will inherit a flag if they should look for a
 565         * child_subreaper process at exit.
 566         */
 567        unsigned int            is_child_subreaper:1;
 568        unsigned int            has_child_subreaper:1;
 569
 570        /* POSIX.1b Interval Timers */
 571        struct list_head posix_timers;
 572
 573        /* ITIMER_REAL timer for the process */
 574        struct hrtimer real_timer;
 575        struct pid *leader_pid;
 576        ktime_t it_real_incr;
 577
 578        /*
 579         * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
 580         * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
 581         * values are defined to 0 and 1 respectively
 582         */
 583        struct cpu_itimer it[2];
 584
 585        /*
 586         * Thread group totals for process CPU timers.
 587         * See thread_group_cputimer(), et al, for details.
 588         */
 589        struct thread_group_cputimer cputimer;
 590
 591        /* Earliest-expiration cache. */
 592        struct task_cputime cputime_expires;
 593
 594        struct list_head cpu_timers[3];
 595
 596        struct pid *tty_old_pgrp;
 597
 598        /* boolean value for session group leader */
 599        int leader;
 600
 601        struct tty_struct *tty; /* NULL if no tty */
 602
 603#ifdef CONFIG_SCHED_AUTOGROUP
 604        struct autogroup *autogroup;
 605#endif
 606        /*
 607         * Cumulative resource counters for dead threads in the group,
 608         * and for reaped dead child processes forked by this group.
 609         * Live threads maintain their own counters and add to these
 610         * in __exit_signal, except for the group leader.
 611         */
 612        cputime_t utime, stime, cutime, cstime;
 613        cputime_t gtime;
 614        cputime_t cgtime;
 615#ifndef CONFIG_VIRT_CPU_ACCOUNTING
 616        cputime_t prev_utime, prev_stime;
 617#endif
 618        unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
 619        unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
 620        unsigned long inblock, oublock, cinblock, coublock;
 621        unsigned long maxrss, cmaxrss;
 622        struct task_io_accounting ioac;
 623
 624        /*
 625         * Cumulative ns of schedule CPU time fo dead threads in the
 626         * group, not including a zombie group leader, (This only differs
 627         * from jiffies_to_ns(utime + stime) if sched_clock uses something
 628         * other than jiffies.)
 629         */
 630        unsigned long long sum_sched_runtime;
 631
 632        /*
 633         * We don't bother to synchronize most readers of this at all,
 634         * because there is no reader checking a limit that actually needs
 635         * to get both rlim_cur and rlim_max atomically, and either one
 636         * alone is a single word that can safely be read normally.
 637         * getrlimit/setrlimit use task_lock(current->group_leader) to
 638         * protect this instead of the siglock, because they really
 639         * have no need to disable irqs.
 640         */
 641        struct rlimit rlim[RLIM_NLIMITS];
 642
 643#ifdef CONFIG_BSD_PROCESS_ACCT
 644        struct pacct_struct pacct;      /* per-process accounting information */
 645#endif
 646#ifdef CONFIG_TASKSTATS
 647        struct taskstats *stats;
 648#endif
 649#ifdef CONFIG_AUDIT
 650        unsigned audit_tty;
 651        struct tty_audit_buf *tty_audit_buf;
 652#endif
 653#ifdef CONFIG_CGROUPS
 654        /*
 655         * group_rwsem prevents new tasks from entering the threadgroup and
 656         * member tasks from exiting,a more specifically, setting of
 657         * PF_EXITING.  fork and exit paths are protected with this rwsem
 658         * using threadgroup_change_begin/end().  Users which require
 659         * threadgroup to remain stable should use threadgroup_[un]lock()
 660         * which also takes care of exec path.  Currently, cgroup is the
 661         * only user.
 662         */
 663        struct rw_semaphore group_rwsem;
 664#endif
 665
 666        int oom_adj;            /* OOM kill score adjustment (bit shift) */
 667        int oom_score_adj;      /* OOM kill score adjustment */
 668        int oom_score_adj_min;  /* OOM kill score adjustment minimum value.
 669                                 * Only settable by CAP_SYS_RESOURCE. */
 670
 671        struct mutex cred_guard_mutex;  /* guard against foreign influences on
 672                                         * credential calculations
 673                                         * (notably. ptrace) */
 674};
 675
 676/* Context switch must be unlocked if interrupts are to be enabled */
 677#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
 678# define __ARCH_WANT_UNLOCKED_CTXSW
 679#endif
 680
 681/*
 682 * Bits in flags field of signal_struct.
 683 */
 684#define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
 685#define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
 686#define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
 687/*
 688 * Pending notifications to parent.
 689 */
 690#define SIGNAL_CLD_STOPPED      0x00000010
 691#define SIGNAL_CLD_CONTINUED    0x00000020
 692#define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
 693
 694#define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
 695
 696/* If true, all threads except ->group_exit_task have pending SIGKILL */
 697static inline int signal_group_exit(const struct signal_struct *sig)
 698{
 699        return  (sig->flags & SIGNAL_GROUP_EXIT) ||
 700                (sig->group_exit_task != NULL);
 701}
 702
 703/*
 704 * Some day this will be a full-fledged user tracking system..
 705 */
 706struct user_struct {
 707        atomic_t __count;       /* reference count */
 708        atomic_t processes;     /* How many processes does this user have? */
 709        atomic_t files;         /* How many open files does this user have? */
 710        atomic_t sigpending;    /* How many pending signals does this user have? */
 711#ifdef CONFIG_INOTIFY_USER
 712        atomic_t inotify_watches; /* How many inotify watches does this user have? */
 713        atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
 714#endif
 715#ifdef CONFIG_FANOTIFY
 716        atomic_t fanotify_listeners;
 717#endif
 718#ifdef CONFIG_EPOLL
 719        atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
 720#endif
 721#ifdef CONFIG_POSIX_MQUEUE
 722        /* protected by mq_lock */
 723        unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
 724#endif
 725        unsigned long locked_shm; /* How many pages of mlocked shm ? */
 726
 727#ifdef CONFIG_KEYS
 728        struct key *uid_keyring;        /* UID specific keyring */
 729        struct key *session_keyring;    /* UID's default session keyring */
 730#endif
 731
 732        /* Hash table maintenance information */
 733        struct hlist_node uidhash_node;
 734        kuid_t uid;
 735
 736#ifdef CONFIG_PERF_EVENTS
 737        atomic_long_t locked_vm;
 738#endif
 739};
 740
 741extern int uids_sysfs_init(void);
 742
 743extern struct user_struct *find_user(kuid_t);
 744
 745extern struct user_struct root_user;
 746#define INIT_USER (&root_user)
 747
 748
 749struct backing_dev_info;
 750struct reclaim_state;
 751
 752#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 753struct sched_info {
 754        /* cumulative counters */
 755        unsigned long pcount;         /* # of times run on this cpu */
 756        unsigned long long run_delay; /* time spent waiting on a runqueue */
 757
 758        /* timestamps */
 759        unsigned long long last_arrival,/* when we last ran on a cpu */
 760                           last_queued; /* when we were last queued to run */
 761};
 762#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
 763
 764#ifdef CONFIG_TASK_DELAY_ACCT
 765struct task_delay_info {
 766        spinlock_t      lock;
 767        unsigned int    flags;  /* Private per-task flags */
 768
 769        /* For each stat XXX, add following, aligned appropriately
 770         *
 771         * struct timespec XXX_start, XXX_end;
 772         * u64 XXX_delay;
 773         * u32 XXX_count;
 774         *
 775         * Atomicity of updates to XXX_delay, XXX_count protected by
 776         * single lock above (split into XXX_lock if contention is an issue).
 777         */
 778
 779        /*
 780         * XXX_count is incremented on every XXX operation, the delay
 781         * associated with the operation is added to XXX_delay.
 782         * XXX_delay contains the accumulated delay time in nanoseconds.
 783         */
 784        struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
 785        u64 blkio_delay;        /* wait for sync block io completion */
 786        u64 swapin_delay;       /* wait for swapin block io completion */
 787        u32 blkio_count;        /* total count of the number of sync block */
 788                                /* io operations performed */
 789        u32 swapin_count;       /* total count of the number of swapin block */
 790                                /* io operations performed */
 791
 792        struct timespec freepages_start, freepages_end;
 793        u64 freepages_delay;    /* wait for memory reclaim */
 794        u32 freepages_count;    /* total count of memory reclaim */
 795};
 796#endif  /* CONFIG_TASK_DELAY_ACCT */
 797
 798static inline int sched_info_on(void)
 799{
 800#ifdef CONFIG_SCHEDSTATS
 801        return 1;
 802#elif defined(CONFIG_TASK_DELAY_ACCT)
 803        extern int delayacct_on;
 804        return delayacct_on;
 805#else
 806        return 0;
 807#endif
 808}
 809
 810enum cpu_idle_type {
 811        CPU_IDLE,
 812        CPU_NOT_IDLE,
 813        CPU_NEWLY_IDLE,
 814        CPU_MAX_IDLE_TYPES
 815};
 816
 817/*
 818 * Increase resolution of nice-level calculations for 64-bit architectures.
 819 * The extra resolution improves shares distribution and load balancing of
 820 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
 821 * hierarchies, especially on larger systems. This is not a user-visible change
 822 * and does not change the user-interface for setting shares/weights.
 823 *
 824 * We increase resolution only if we have enough bits to allow this increased
 825 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
 826 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
 827 * increased costs.
 828 */
 829#if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
 830# define SCHED_LOAD_RESOLUTION  10
 831# define scale_load(w)          ((w) << SCHED_LOAD_RESOLUTION)
 832# define scale_load_down(w)     ((w) >> SCHED_LOAD_RESOLUTION)
 833#else
 834# define SCHED_LOAD_RESOLUTION  0
 835# define scale_load(w)          (w)
 836# define scale_load_down(w)     (w)
 837#endif
 838
 839#define SCHED_LOAD_SHIFT        (10 + SCHED_LOAD_RESOLUTION)
 840#define SCHED_LOAD_SCALE        (1L << SCHED_LOAD_SHIFT)
 841
 842/*
 843 * Increase resolution of cpu_power calculations
 844 */
 845#define SCHED_POWER_SHIFT       10
 846#define SCHED_POWER_SCALE       (1L << SCHED_POWER_SHIFT)
 847
 848/*
 849 * sched-domains (multiprocessor balancing) declarations:
 850 */
 851#ifdef CONFIG_SMP
 852#define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
 853#define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
 854#define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
 855#define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
 856#define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
 857#define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
 858#define SD_PREFER_LOCAL         0x0040  /* Prefer to keep tasks local to this domain */
 859#define SD_SHARE_CPUPOWER       0x0080  /* Domain members share cpu power */
 860#define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
 861#define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
 862#define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
 863#define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
 864#define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
 865
 866extern int __weak arch_sd_sibiling_asym_packing(void);
 867
 868struct sched_group_power {
 869        atomic_t ref;
 870        /*
 871         * CPU power of this group, SCHED_LOAD_SCALE being max power for a
 872         * single CPU.
 873         */
 874        unsigned int power, power_orig;
 875        unsigned long next_update;
 876        /*
 877         * Number of busy cpus in this group.
 878         */
 879        atomic_t nr_busy_cpus;
 880
 881        unsigned long cpumask[0]; /* iteration mask */
 882};
 883
 884struct sched_group {
 885        struct sched_group *next;       /* Must be a circular list */
 886        atomic_t ref;
 887
 888        unsigned int group_weight;
 889        struct sched_group_power *sgp;
 890
 891        /*
 892         * The CPUs this group covers.
 893         *
 894         * NOTE: this field is variable length. (Allocated dynamically
 895         * by attaching extra space to the end of the structure,
 896         * depending on how many CPUs the kernel has booted up with)
 897         */
 898        unsigned long cpumask[0];
 899};
 900
 901static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
 902{
 903        return to_cpumask(sg->cpumask);
 904}
 905
 906/*
 907 * cpumask masking which cpus in the group are allowed to iterate up the domain
 908 * tree.
 909 */
 910static inline struct cpumask *sched_group_mask(struct sched_group *sg)
 911{
 912        return to_cpumask(sg->sgp->cpumask);
 913}
 914
 915/**
 916 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
 917 * @group: The group whose first cpu is to be returned.
 918 */
 919static inline unsigned int group_first_cpu(struct sched_group *group)
 920{
 921        return cpumask_first(sched_group_cpus(group));
 922}
 923
 924struct sched_domain_attr {
 925        int relax_domain_level;
 926};
 927
 928#define SD_ATTR_INIT    (struct sched_domain_attr) {    \
 929        .relax_domain_level = -1,                       \
 930}
 931
 932extern int sched_domain_level_max;
 933
 934struct sched_domain {
 935        /* These fields must be setup */
 936        struct sched_domain *parent;    /* top domain must be null terminated */
 937        struct sched_domain *child;     /* bottom domain must be null terminated */
 938        struct sched_group *groups;     /* the balancing groups of the domain */
 939        unsigned long min_interval;     /* Minimum balance interval ms */
 940        unsigned long max_interval;     /* Maximum balance interval ms */
 941        unsigned int busy_factor;       /* less balancing by factor if busy */
 942        unsigned int imbalance_pct;     /* No balance until over watermark */
 943        unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
 944        unsigned int busy_idx;
 945        unsigned int idle_idx;
 946        unsigned int newidle_idx;
 947        unsigned int wake_idx;
 948        unsigned int forkexec_idx;
 949        unsigned int smt_gain;
 950        int flags;                      /* See SD_* */
 951        int level;
 952
 953        /* Runtime fields. */
 954        unsigned long last_balance;     /* init to jiffies. units in jiffies */
 955        unsigned int balance_interval;  /* initialise to 1. units in ms. */
 956        unsigned int nr_balance_failed; /* initialise to 0 */
 957
 958        u64 last_update;
 959
 960#ifdef CONFIG_SCHEDSTATS
 961        /* load_balance() stats */
 962        unsigned int lb_count[CPU_MAX_IDLE_TYPES];
 963        unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
 964        unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
 965        unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
 966        unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
 967        unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
 968        unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
 969        unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
 970
 971        /* Active load balancing */
 972        unsigned int alb_count;
 973        unsigned int alb_failed;
 974        unsigned int alb_pushed;
 975
 976        /* SD_BALANCE_EXEC stats */
 977        unsigned int sbe_count;
 978        unsigned int sbe_balanced;
 979        unsigned int sbe_pushed;
 980
 981        /* SD_BALANCE_FORK stats */
 982        unsigned int sbf_count;
 983        unsigned int sbf_balanced;
 984        unsigned int sbf_pushed;
 985
 986        /* try_to_wake_up() stats */
 987        unsigned int ttwu_wake_remote;
 988        unsigned int ttwu_move_affine;
 989        unsigned int ttwu_move_balance;
 990#endif
 991#ifdef CONFIG_SCHED_DEBUG
 992        char *name;
 993#endif
 994        union {
 995                void *private;          /* used during construction */
 996                struct rcu_head rcu;    /* used during destruction */
 997        };
 998
 999        unsigned int span_weight;
1000        /*
1001         * Span of all CPUs in this domain.
1002         *
1003         * NOTE: this field is variable length. (Allocated dynamically
1004         * by attaching extra space to the end of the structure,
1005         * depending on how many CPUs the kernel has booted up with)
1006         */
1007        unsigned long span[0];
1008};
1009
1010static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1011{
1012        return to_cpumask(sd->span);
1013}
1014
1015extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1016                                    struct sched_domain_attr *dattr_new);
1017
1018/* Allocate an array of sched domains, for partition_sched_domains(). */
1019cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1020void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1021
1022/* Test a flag in parent sched domain */
1023static inline int test_sd_parent(struct sched_domain *sd, int flag)
1024{
1025        if (sd->parent && (sd->parent->flags & flag))
1026                return 1;
1027
1028        return 0;
1029}
1030
1031unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1032unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1033
1034bool cpus_share_cache(int this_cpu, int that_cpu);
1035
1036#else /* CONFIG_SMP */
1037
1038struct sched_domain_attr;
1039
1040static inline void
1041partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1042                        struct sched_domain_attr *dattr_new)
1043{
1044}
1045
1046static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1047{
1048        return true;
1049}
1050
1051#endif  /* !CONFIG_SMP */
1052
1053
1054struct io_context;                      /* See blkdev.h */
1055
1056
1057#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1058extern void prefetch_stack(struct task_struct *t);
1059#else
1060static inline void prefetch_stack(struct task_struct *t) { }
1061#endif
1062
1063struct audit_context;           /* See audit.c */
1064struct mempolicy;
1065struct pipe_inode_info;
1066struct uts_namespace;
1067
1068struct rq;
1069struct sched_domain;
1070
1071/*
1072 * wake flags
1073 */
1074#define WF_SYNC         0x01            /* waker goes to sleep after wakup */
1075#define WF_FORK         0x02            /* child wakeup after fork */
1076#define WF_MIGRATED     0x04            /* internal use, task got migrated */
1077
1078#define ENQUEUE_WAKEUP          1
1079#define ENQUEUE_HEAD            2
1080#ifdef CONFIG_SMP
1081#define ENQUEUE_WAKING          4       /* sched_class::task_waking was called */
1082#else
1083#define ENQUEUE_WAKING          0
1084#endif
1085
1086#define DEQUEUE_SLEEP           1
1087
1088struct sched_class {
1089        const struct sched_class *next;
1090
1091        void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1092        void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1093        void (*yield_task) (struct rq *rq);
1094        bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1095
1096        void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1097
1098        struct task_struct * (*pick_next_task) (struct rq *rq);
1099        void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1100
1101#ifdef CONFIG_SMP
1102        int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1103
1104        void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1105        void (*post_schedule) (struct rq *this_rq);
1106        void (*task_waking) (struct task_struct *task);
1107        void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1108
1109        void (*set_cpus_allowed)(struct task_struct *p,
1110                                 const struct cpumask *newmask);
1111
1112        void (*rq_online)(struct rq *rq);
1113        void (*rq_offline)(struct rq *rq);
1114#endif
1115
1116        void (*set_curr_task) (struct rq *rq);
1117        void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1118        void (*task_fork) (struct task_struct *p);
1119
1120        void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1121        void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1122        void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1123                             int oldprio);
1124
1125        unsigned int (*get_rr_interval) (struct rq *rq,
1126                                         struct task_struct *task);
1127
1128#ifdef CONFIG_FAIR_GROUP_SCHED
1129        void (*task_move_group) (struct task_struct *p, int on_rq);
1130#endif
1131};
1132
1133struct load_weight {
1134        unsigned long weight, inv_weight;
1135};
1136
1137#ifdef CONFIG_SCHEDSTATS
1138struct sched_statistics {
1139        u64                     wait_start;
1140        u64                     wait_max;
1141        u64                     wait_count;
1142        u64                     wait_sum;
1143        u64                     iowait_count;
1144        u64                     iowait_sum;
1145
1146        u64                     sleep_start;
1147        u64                     sleep_max;
1148        s64                     sum_sleep_runtime;
1149
1150        u64                     block_start;
1151        u64                     block_max;
1152        u64                     exec_max;
1153        u64                     slice_max;
1154
1155        u64                     nr_migrations_cold;
1156        u64                     nr_failed_migrations_affine;
1157        u64                     nr_failed_migrations_running;
1158        u64                     nr_failed_migrations_hot;
1159        u64                     nr_forced_migrations;
1160
1161        u64                     nr_wakeups;
1162        u64                     nr_wakeups_sync;
1163        u64                     nr_wakeups_migrate;
1164        u64                     nr_wakeups_local;
1165        u64                     nr_wakeups_remote;
1166        u64                     nr_wakeups_affine;
1167        u64                     nr_wakeups_affine_attempts;
1168        u64                     nr_wakeups_passive;
1169        u64                     nr_wakeups_idle;
1170};
1171#endif
1172
1173struct sched_entity {
1174        struct load_weight      load;           /* for load-balancing */
1175        struct rb_node          run_node;
1176        struct list_head        group_node;
1177        unsigned int            on_rq;
1178
1179        u64                     exec_start;
1180        u64                     sum_exec_runtime;
1181        u64                     vruntime;
1182        u64                     prev_sum_exec_runtime;
1183
1184        u64                     nr_migrations;
1185
1186#ifdef CONFIG_SCHEDSTATS
1187        struct sched_statistics statistics;
1188#endif
1189
1190#ifdef CONFIG_FAIR_GROUP_SCHED
1191        struct sched_entity     *parent;
1192        /* rq on which this entity is (to be) queued: */
1193        struct cfs_rq           *cfs_rq;
1194        /* rq "owned" by this entity/group: */
1195        struct cfs_rq           *my_q;
1196#endif
1197};
1198
1199struct sched_rt_entity {
1200        struct list_head run_list;
1201        unsigned long timeout;
1202        unsigned int time_slice;
1203
1204        struct sched_rt_entity *back;
1205#ifdef CONFIG_RT_GROUP_SCHED
1206        struct sched_rt_entity  *parent;
1207        /* rq on which this entity is (to be) queued: */
1208        struct rt_rq            *rt_rq;
1209        /* rq "owned" by this entity/group: */
1210        struct rt_rq            *my_q;
1211#endif
1212};
1213
1214/*
1215 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1216 * Timeslices get refilled after they expire.
1217 */
1218#define RR_TIMESLICE            (100 * HZ / 1000)
1219
1220struct rcu_node;
1221
1222enum perf_event_task_context {
1223        perf_invalid_context = -1,
1224        perf_hw_context = 0,
1225        perf_sw_context,
1226        perf_nr_task_contexts,
1227};
1228
1229struct task_struct {
1230        volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1231        void *stack;
1232        atomic_t usage;
1233        unsigned int flags;     /* per process flags, defined below */
1234        unsigned int ptrace;
1235
1236#ifdef CONFIG_SMP
1237        struct llist_node wake_entry;
1238        int on_cpu;
1239#endif
1240        int on_rq;
1241
1242        int prio, static_prio, normal_prio;
1243        unsigned int rt_priority;
1244        const struct sched_class *sched_class;
1245        struct sched_entity se;
1246        struct sched_rt_entity rt;
1247
1248#ifdef CONFIG_PREEMPT_NOTIFIERS
1249        /* list of struct preempt_notifier: */
1250        struct hlist_head preempt_notifiers;
1251#endif
1252
1253        /*
1254         * fpu_counter contains the number of consecutive context switches
1255         * that the FPU is used. If this is over a threshold, the lazy fpu
1256         * saving becomes unlazy to save the trap. This is an unsigned char
1257         * so that after 256 times the counter wraps and the behavior turns
1258         * lazy again; this to deal with bursty apps that only use FPU for
1259         * a short time
1260         */
1261        unsigned char fpu_counter;
1262#ifdef CONFIG_BLK_DEV_IO_TRACE
1263        unsigned int btrace_seq;
1264#endif
1265
1266        unsigned int policy;
1267        int nr_cpus_allowed;
1268        cpumask_t cpus_allowed;
1269
1270#ifdef CONFIG_PREEMPT_RCU
1271        int rcu_read_lock_nesting;
1272        char rcu_read_unlock_special;
1273        struct list_head rcu_node_entry;
1274#endif /* #ifdef CONFIG_PREEMPT_RCU */
1275#ifdef CONFIG_TREE_PREEMPT_RCU
1276        struct rcu_node *rcu_blocked_node;
1277#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1278#ifdef CONFIG_RCU_BOOST
1279        struct rt_mutex *rcu_boost_mutex;
1280#endif /* #ifdef CONFIG_RCU_BOOST */
1281
1282#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1283        struct sched_info sched_info;
1284#endif
1285
1286        struct list_head tasks;
1287#ifdef CONFIG_SMP
1288        struct plist_node pushable_tasks;
1289#endif
1290
1291        struct mm_struct *mm, *active_mm;
1292#ifdef CONFIG_COMPAT_BRK
1293        unsigned brk_randomized:1;
1294#endif
1295#if defined(SPLIT_RSS_COUNTING)
1296        struct task_rss_stat    rss_stat;
1297#endif
1298/* task state */
1299        int exit_state;
1300        int exit_code, exit_signal;
1301        int pdeath_signal;  /*  The signal sent when the parent dies  */
1302        unsigned int jobctl;    /* JOBCTL_*, siglock protected */
1303        /* ??? */
1304        unsigned int personality;
1305        unsigned did_exec:1;
1306        unsigned in_execve:1;   /* Tell the LSMs that the process is doing an
1307                                 * execve */
1308        unsigned in_iowait:1;
1309
1310        /* task may not gain privileges */
1311        unsigned no_new_privs:1;
1312
1313        /* Revert to default priority/policy when forking */
1314        unsigned sched_reset_on_fork:1;
1315        unsigned sched_contributes_to_load:1;
1316
1317        pid_t pid;
1318        pid_t tgid;
1319
1320#ifdef CONFIG_CC_STACKPROTECTOR
1321        /* Canary value for the -fstack-protector gcc feature */
1322        unsigned long stack_canary;
1323#endif
1324        /*
1325         * pointers to (original) parent process, youngest child, younger sibling,
1326         * older sibling, respectively.  (p->father can be replaced with
1327         * p->real_parent->pid)
1328         */
1329        struct task_struct __rcu *real_parent; /* real parent process */
1330        struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1331        /*
1332         * children/sibling forms the list of my natural children
1333         */
1334        struct list_head children;      /* list of my children */
1335        struct list_head sibling;       /* linkage in my parent's children list */
1336        struct task_struct *group_leader;       /* threadgroup leader */
1337
1338        /*
1339         * ptraced is the list of tasks this task is using ptrace on.
1340         * This includes both natural children and PTRACE_ATTACH targets.
1341         * p->ptrace_entry is p's link on the p->parent->ptraced list.
1342         */
1343        struct list_head ptraced;
1344        struct list_head ptrace_entry;
1345
1346        /* PID/PID hash table linkage. */
1347        struct pid_link pids[PIDTYPE_MAX];
1348        struct list_head thread_group;
1349
1350        struct completion *vfork_done;          /* for vfork() */
1351        int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1352        int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1353
1354        cputime_t utime, stime, utimescaled, stimescaled;
1355        cputime_t gtime;
1356#ifndef CONFIG_VIRT_CPU_ACCOUNTING
1357        cputime_t prev_utime, prev_stime;
1358#endif
1359        unsigned long nvcsw, nivcsw; /* context switch counts */
1360        struct timespec start_time;             /* monotonic time */
1361        struct timespec real_start_time;        /* boot based time */
1362/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1363        unsigned long min_flt, maj_flt;
1364
1365        struct task_cputime cputime_expires;
1366        struct list_head cpu_timers[3];
1367
1368/* process credentials */
1369        const struct cred __rcu *real_cred; /* objective and real subjective task
1370                                         * credentials (COW) */
1371        const struct cred __rcu *cred;  /* effective (overridable) subjective task
1372                                         * credentials (COW) */
1373        char comm[TASK_COMM_LEN]; /* executable name excluding path
1374                                     - access with [gs]et_task_comm (which lock
1375                                       it with task_lock())
1376                                     - initialized normally by setup_new_exec */
1377/* file system info */
1378        int link_count, total_link_count;
1379#ifdef CONFIG_SYSVIPC
1380/* ipc stuff */
1381        struct sysv_sem sysvsem;
1382#endif
1383#ifdef CONFIG_DETECT_HUNG_TASK
1384/* hung task detection */
1385        unsigned long last_switch_count;
1386#endif
1387/* CPU-specific state of this task */
1388        struct thread_struct thread;
1389/* filesystem information */
1390        struct fs_struct *fs;
1391/* open file information */
1392        struct files_struct *files;
1393/* namespaces */
1394        struct nsproxy *nsproxy;
1395/* signal handlers */
1396        struct signal_struct *signal;
1397        struct sighand_struct *sighand;
1398
1399        sigset_t blocked, real_blocked;
1400        sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1401        struct sigpending pending;
1402
1403        unsigned long sas_ss_sp;
1404        size_t sas_ss_size;
1405        int (*notifier)(void *priv);
1406        void *notifier_data;
1407        sigset_t *notifier_mask;
1408        struct hlist_head task_works;
1409
1410        struct audit_context *audit_context;
1411#ifdef CONFIG_AUDITSYSCALL
1412        uid_t loginuid;
1413        unsigned int sessionid;
1414#endif
1415        struct seccomp seccomp;
1416
1417/* Thread group tracking */
1418        u32 parent_exec_id;
1419        u32 self_exec_id;
1420/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1421 * mempolicy */
1422        spinlock_t alloc_lock;
1423
1424        /* Protection of the PI data structures: */
1425        raw_spinlock_t pi_lock;
1426
1427#ifdef CONFIG_RT_MUTEXES
1428        /* PI waiters blocked on a rt_mutex held by this task */
1429        struct plist_head pi_waiters;
1430        /* Deadlock detection and priority inheritance handling */
1431        struct rt_mutex_waiter *pi_blocked_on;
1432#endif
1433
1434#ifdef CONFIG_DEBUG_MUTEXES
1435        /* mutex deadlock detection */
1436        struct mutex_waiter *blocked_on;
1437#endif
1438#ifdef CONFIG_TRACE_IRQFLAGS
1439        unsigned int irq_events;
1440        unsigned long hardirq_enable_ip;
1441        unsigned long hardirq_disable_ip;
1442        unsigned int hardirq_enable_event;
1443        unsigned int hardirq_disable_event;
1444        int hardirqs_enabled;
1445        int hardirq_context;
1446        unsigned long softirq_disable_ip;
1447        unsigned long softirq_enable_ip;
1448        unsigned int softirq_disable_event;
1449        unsigned int softirq_enable_event;
1450        int softirqs_enabled;
1451        int softirq_context;
1452#endif
1453#ifdef CONFIG_LOCKDEP
1454# define MAX_LOCK_DEPTH 48UL
1455        u64 curr_chain_key;
1456        int lockdep_depth;
1457        unsigned int lockdep_recursion;
1458        struct held_lock held_locks[MAX_LOCK_DEPTH];
1459        gfp_t lockdep_reclaim_gfp;
1460#endif
1461
1462/* journalling filesystem info */
1463        void *journal_info;
1464
1465/* stacked block device info */
1466        struct bio_list *bio_list;
1467
1468#ifdef CONFIG_BLOCK
1469/* stack plugging */
1470        struct blk_plug *plug;
1471#endif
1472
1473/* VM state */
1474        struct reclaim_state *reclaim_state;
1475
1476        struct backing_dev_info *backing_dev_info;
1477
1478        struct io_context *io_context;
1479
1480        unsigned long ptrace_message;
1481        siginfo_t *last_siginfo; /* For ptrace use.  */
1482        struct task_io_accounting ioac;
1483#if defined(CONFIG_TASK_XACCT)
1484        u64 acct_rss_mem1;      /* accumulated rss usage */
1485        u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1486        cputime_t acct_timexpd; /* stime + utime since last update */
1487#endif
1488#ifdef CONFIG_CPUSETS
1489        nodemask_t mems_allowed;        /* Protected by alloc_lock */
1490        seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1491        int cpuset_mem_spread_rotor;
1492        int cpuset_slab_spread_rotor;
1493#endif
1494#ifdef CONFIG_CGROUPS
1495        /* Control Group info protected by css_set_lock */
1496        struct css_set __rcu *cgroups;
1497        /* cg_list protected by css_set_lock and tsk->alloc_lock */
1498        struct list_head cg_list;
1499#endif
1500#ifdef CONFIG_FUTEX
1501        struct robust_list_head __user *robust_list;
1502#ifdef CONFIG_COMPAT
1503        struct compat_robust_list_head __user *compat_robust_list;
1504#endif
1505        struct list_head pi_state_list;
1506        struct futex_pi_state *pi_state_cache;
1507#endif
1508#ifdef CONFIG_PERF_EVENTS
1509        struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1510        struct mutex perf_event_mutex;
1511        struct list_head perf_event_list;
1512#endif
1513#ifdef CONFIG_NUMA
1514        struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1515        short il_next;
1516        short pref_node_fork;
1517#endif
1518        struct rcu_head rcu;
1519
1520        /*
1521         * cache last used pipe for splice
1522         */
1523        struct pipe_inode_info *splice_pipe;
1524#ifdef  CONFIG_TASK_DELAY_ACCT
1525        struct task_delay_info *delays;
1526#endif
1527#ifdef CONFIG_FAULT_INJECTION
1528        int make_it_fail;
1529#endif
1530        /*
1531         * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1532         * balance_dirty_pages() for some dirty throttling pause
1533         */
1534        int nr_dirtied;
1535        int nr_dirtied_pause;
1536        unsigned long dirty_paused_when; /* start of a write-and-pause period */
1537
1538#ifdef CONFIG_LATENCYTOP
1539        int latency_record_count;
1540        struct latency_record latency_record[LT_SAVECOUNT];
1541#endif
1542        /*
1543         * time slack values; these are used to round up poll() and
1544         * select() etc timeout values. These are in nanoseconds.
1545         */
1546        unsigned long timer_slack_ns;
1547        unsigned long default_timer_slack_ns;
1548
1549        struct list_head        *scm_work_list;
1550#ifdef CONFIG_FUNCTION_GRAPH_TRACER
1551        /* Index of current stored address in ret_stack */
1552        int curr_ret_stack;
1553        /* Stack of return addresses for return function tracing */
1554        struct ftrace_ret_stack *ret_stack;
1555        /* time stamp for last schedule */
1556        unsigned long long ftrace_timestamp;
1557        /*
1558         * Number of functions that haven't been traced
1559         * because of depth overrun.
1560         */
1561        atomic_t trace_overrun;
1562        /* Pause for the tracing */
1563        atomic_t tracing_graph_pause;
1564#endif
1565#ifdef CONFIG_TRACING
1566        /* state flags for use by tracers */
1567        unsigned long trace;
1568        /* bitmask and counter of trace recursion */
1569        unsigned long trace_recursion;
1570#endif /* CONFIG_TRACING */
1571#ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1572        struct memcg_batch_info {
1573                int do_batch;   /* incremented when batch uncharge started */
1574                struct mem_cgroup *memcg; /* target memcg of uncharge */
1575                unsigned long nr_pages; /* uncharged usage */
1576                unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1577        } memcg_batch;
1578#endif
1579#ifdef CONFIG_HAVE_HW_BREAKPOINT
1580        atomic_t ptrace_bp_refcnt;
1581#endif
1582#ifdef CONFIG_UPROBES
1583        struct uprobe_task *utask;
1584        int uprobe_srcu_id;
1585#endif
1586};
1587
1588/* Future-safe accessor for struct task_struct's cpus_allowed. */
1589#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1590
1591/*
1592 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1593 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1594 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1595 * values are inverted: lower p->prio value means higher priority.
1596 *
1597 * The MAX_USER_RT_PRIO value allows the actual maximum
1598 * RT priority to be separate from the value exported to
1599 * user-space.  This allows kernel threads to set their
1600 * priority to a value higher than any user task. Note:
1601 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1602 */
1603
1604#define MAX_USER_RT_PRIO        100
1605#define MAX_RT_PRIO             MAX_USER_RT_PRIO
1606
1607#define MAX_PRIO                (MAX_RT_PRIO + 40)
1608#define DEFAULT_PRIO            (MAX_RT_PRIO + 20)
1609
1610static inline int rt_prio(int prio)
1611{
1612        if (unlikely(prio < MAX_RT_PRIO))
1613                return 1;
1614        return 0;
1615}
1616
1617static inline int rt_task(struct task_struct *p)
1618{
1619        return rt_prio(p->prio);
1620}
1621
1622static inline struct pid *task_pid(struct task_struct *task)
1623{
1624        return task->pids[PIDTYPE_PID].pid;
1625}
1626
1627static inline struct pid *task_tgid(struct task_struct *task)
1628{
1629        return task->group_leader->pids[PIDTYPE_PID].pid;
1630}
1631
1632/*
1633 * Without tasklist or rcu lock it is not safe to dereference
1634 * the result of task_pgrp/task_session even if task == current,
1635 * we can race with another thread doing sys_setsid/sys_setpgid.
1636 */
1637static inline struct pid *task_pgrp(struct task_struct *task)
1638{
1639        return task->group_leader->pids[PIDTYPE_PGID].pid;
1640}
1641
1642static inline struct pid *task_session(struct task_struct *task)
1643{
1644        return task->group_leader->pids[PIDTYPE_SID].pid;
1645}
1646
1647struct pid_namespace;
1648
1649/*
1650 * the helpers to get the task's different pids as they are seen
1651 * from various namespaces
1652 *
1653 * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1654 * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1655 *                     current.
1656 * task_xid_nr_ns()  : id seen from the ns specified;
1657 *
1658 * set_task_vxid()   : assigns a virtual id to a task;
1659 *
1660 * see also pid_nr() etc in include/linux/pid.h
1661 */
1662pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1663                        struct pid_namespace *ns);
1664
1665static inline pid_t task_pid_nr(struct task_struct *tsk)
1666{
1667        return tsk->pid;
1668}
1669
1670static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1671                                        struct pid_namespace *ns)
1672{
1673        return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1674}
1675
1676static inline pid_t task_pid_vnr(struct task_struct *tsk)
1677{
1678        return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1679}
1680
1681
1682static inline pid_t task_tgid_nr(struct task_struct *tsk)
1683{
1684        return tsk->tgid;
1685}
1686
1687pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1688
1689static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1690{
1691        return pid_vnr(task_tgid(tsk));
1692}
1693
1694
1695static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1696                                        struct pid_namespace *ns)
1697{
1698        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1699}
1700
1701static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1702{
1703        return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1704}
1705
1706
1707static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1708                                        struct pid_namespace *ns)
1709{
1710        return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1711}
1712
1713static inline pid_t task_session_vnr(struct task_struct *tsk)
1714{
1715        return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1716}
1717
1718/* obsolete, do not use */
1719static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1720{
1721        return task_pgrp_nr_ns(tsk, &init_pid_ns);
1722}
1723
1724/**
1725 * pid_alive - check that a task structure is not stale
1726 * @p: Task structure to be checked.
1727 *
1728 * Test if a process is not yet dead (at most zombie state)
1729 * If pid_alive fails, then pointers within the task structure
1730 * can be stale and must not be dereferenced.
1731 */
1732static inline int pid_alive(struct task_struct *p)
1733{
1734        return p->pids[PIDTYPE_PID].pid != NULL;
1735}
1736
1737/**
1738 * is_global_init - check if a task structure is init
1739 * @tsk: Task structure to be checked.
1740 *
1741 * Check if a task structure is the first user space task the kernel created.
1742 */
1743static inline int is_global_init(struct task_struct *tsk)
1744{
1745        return tsk->pid == 1;
1746}
1747
1748/*
1749 * is_container_init:
1750 * check whether in the task is init in its own pid namespace.
1751 */
1752extern int is_container_init(struct task_struct *tsk);
1753
1754extern struct pid *cad_pid;
1755
1756extern void free_task(struct task_struct *tsk);
1757#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1758
1759extern void __put_task_struct(struct task_struct *t);
1760
1761static inline void put_task_struct(struct task_struct *t)
1762{
1763        if (atomic_dec_and_test(&t->usage))
1764                __put_task_struct(t);
1765}
1766
1767extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1768extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1769
1770/*
1771 * Per process flags
1772 */
1773#define PF_EXITING      0x00000004      /* getting shut down */
1774#define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
1775#define PF_VCPU         0x00000010      /* I'm a virtual CPU */
1776#define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
1777#define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
1778#define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1779#define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
1780#define PF_DUMPCORE     0x00000200      /* dumped core */
1781#define PF_SIGNALED     0x00000400      /* killed by a signal */
1782#define PF_MEMALLOC     0x00000800      /* Allocating memory */
1783#define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
1784#define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
1785#define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
1786#define PF_FROZEN       0x00010000      /* frozen for system suspend */
1787#define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
1788#define PF_KSWAPD       0x00040000      /* I am kswapd */
1789#define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
1790#define PF_KTHREAD      0x00200000      /* I am a kernel thread */
1791#define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
1792#define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
1793#define PF_SPREAD_PAGE  0x01000000      /* Spread page cache over cpuset */
1794#define PF_SPREAD_SLAB  0x02000000      /* Spread some slab caches over cpuset */
1795#define PF_THREAD_BOUND 0x04000000      /* Thread bound to specific cpu */
1796#define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1797#define PF_MEMPOLICY    0x10000000      /* Non-default NUMA mempolicy */
1798#define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
1799#define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
1800
1801/*
1802 * Only the _current_ task can read/write to tsk->flags, but other
1803 * tasks can access tsk->flags in readonly mode for example
1804 * with tsk_used_math (like during threaded core dumping).
1805 * There is however an exception to this rule during ptrace
1806 * or during fork: the ptracer task is allowed to write to the
1807 * child->flags of its traced child (same goes for fork, the parent
1808 * can write to the child->flags), because we're guaranteed the
1809 * child is not running and in turn not changing child->flags
1810 * at the same time the parent does it.
1811 */
1812#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1813#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1814#define clear_used_math() clear_stopped_child_used_math(current)
1815#define set_used_math() set_stopped_child_used_math(current)
1816#define conditional_stopped_child_used_math(condition, child) \
1817        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1818#define conditional_used_math(condition) \
1819        conditional_stopped_child_used_math(condition, current)
1820#define copy_to_stopped_child_used_math(child) \
1821        do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1822/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1823#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1824#define used_math() tsk_used_math(current)
1825
1826/*
1827 * task->jobctl flags
1828 */
1829#define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
1830
1831#define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
1832#define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
1833#define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
1834#define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
1835#define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
1836#define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
1837#define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
1838
1839#define JOBCTL_STOP_DEQUEUED    (1 << JOBCTL_STOP_DEQUEUED_BIT)
1840#define JOBCTL_STOP_PENDING     (1 << JOBCTL_STOP_PENDING_BIT)
1841#define JOBCTL_STOP_CONSUME     (1 << JOBCTL_STOP_CONSUME_BIT)
1842#define JOBCTL_TRAP_STOP        (1 << JOBCTL_TRAP_STOP_BIT)
1843#define JOBCTL_TRAP_NOTIFY      (1 << JOBCTL_TRAP_NOTIFY_BIT)
1844#define JOBCTL_TRAPPING         (1 << JOBCTL_TRAPPING_BIT)
1845#define JOBCTL_LISTENING        (1 << JOBCTL_LISTENING_BIT)
1846
1847#define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1848#define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1849
1850extern bool task_set_jobctl_pending(struct task_struct *task,
1851                                    unsigned int mask);
1852extern void task_clear_jobctl_trapping(struct task_struct *task);
1853extern void task_clear_jobctl_pending(struct task_struct *task,
1854                                      unsigned int mask);
1855
1856#ifdef CONFIG_PREEMPT_RCU
1857
1858#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1859#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1860
1861static inline void rcu_copy_process(struct task_struct *p)
1862{
1863        p->rcu_read_lock_nesting = 0;
1864        p->rcu_read_unlock_special = 0;
1865#ifdef CONFIG_TREE_PREEMPT_RCU
1866        p->rcu_blocked_node = NULL;
1867#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1868#ifdef CONFIG_RCU_BOOST
1869        p->rcu_boost_mutex = NULL;
1870#endif /* #ifdef CONFIG_RCU_BOOST */
1871        INIT_LIST_HEAD(&p->rcu_node_entry);
1872}
1873
1874#else
1875
1876static inline void rcu_copy_process(struct task_struct *p)
1877{
1878}
1879
1880#endif
1881
1882#ifdef CONFIG_SMP
1883extern void do_set_cpus_allowed(struct task_struct *p,
1884                               const struct cpumask *new_mask);
1885
1886extern int set_cpus_allowed_ptr(struct task_struct *p,
1887                                const struct cpumask *new_mask);
1888#else
1889static inline void do_set_cpus_allowed(struct task_struct *p,
1890                                      const struct cpumask *new_mask)
1891{
1892}
1893static inline int set_cpus_allowed_ptr(struct task_struct *p,
1894                                       const struct cpumask *new_mask)
1895{
1896        if (!cpumask_test_cpu(0, new_mask))
1897                return -EINVAL;
1898        return 0;
1899}
1900#endif
1901
1902#ifdef CONFIG_NO_HZ
1903void calc_load_enter_idle(void);
1904void calc_load_exit_idle(void);
1905#else
1906static inline void calc_load_enter_idle(void) { }
1907static inline void calc_load_exit_idle(void) { }
1908#endif /* CONFIG_NO_HZ */
1909
1910#ifndef CONFIG_CPUMASK_OFFSTACK
1911static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1912{
1913        return set_cpus_allowed_ptr(p, &new_mask);
1914}
1915#endif
1916
1917/*
1918 * Do not use outside of architecture code which knows its limitations.
1919 *
1920 * sched_clock() has no promise of monotonicity or bounded drift between
1921 * CPUs, use (which you should not) requires disabling IRQs.
1922 *
1923 * Please use one of the three interfaces below.
1924 */
1925extern unsigned long long notrace sched_clock(void);
1926/*
1927 * See the comment in kernel/sched/clock.c
1928 */
1929extern u64 cpu_clock(int cpu);
1930extern u64 local_clock(void);
1931extern u64 sched_clock_cpu(int cpu);
1932
1933
1934extern void sched_clock_init(void);
1935
1936#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1937static inline void sched_clock_tick(void)
1938{
1939}
1940
1941static inline void sched_clock_idle_sleep_event(void)
1942{
1943}
1944
1945static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1946{
1947}
1948#else
1949/*
1950 * Architectures can set this to 1 if they have specified
1951 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1952 * but then during bootup it turns out that sched_clock()
1953 * is reliable after all:
1954 */
1955extern int sched_clock_stable;
1956
1957extern void sched_clock_tick(void);
1958extern void sched_clock_idle_sleep_event(void);
1959extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1960#endif
1961
1962#ifdef CONFIG_IRQ_TIME_ACCOUNTING
1963/*
1964 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1965 * The reason for this explicit opt-in is not to have perf penalty with
1966 * slow sched_clocks.
1967 */
1968extern void enable_sched_clock_irqtime(void);
1969extern void disable_sched_clock_irqtime(void);
1970#else
1971static inline void enable_sched_clock_irqtime(void) {}
1972static inline void disable_sched_clock_irqtime(void) {}
1973#endif
1974
1975extern unsigned long long
1976task_sched_runtime(struct task_struct *task);
1977
1978/* sched_exec is called by processes performing an exec */
1979#ifdef CONFIG_SMP
1980extern void sched_exec(void);
1981#else
1982#define sched_exec()   {}
1983#endif
1984
1985extern void sched_clock_idle_sleep_event(void);
1986extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1987
1988#ifdef CONFIG_HOTPLUG_CPU
1989extern void idle_task_exit(void);
1990#else
1991static inline void idle_task_exit(void) {}
1992#endif
1993
1994#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1995extern void wake_up_idle_cpu(int cpu);
1996#else
1997static inline void wake_up_idle_cpu(int cpu) { }
1998#endif
1999
2000extern unsigned int sysctl_sched_latency;
2001extern unsigned int sysctl_sched_min_granularity;
2002extern unsigned int sysctl_sched_wakeup_granularity;
2003extern unsigned int sysctl_sched_child_runs_first;
2004
2005enum sched_tunable_scaling {
2006        SCHED_TUNABLESCALING_NONE,
2007        SCHED_TUNABLESCALING_LOG,
2008        SCHED_TUNABLESCALING_LINEAR,
2009        SCHED_TUNABLESCALING_END,
2010};
2011extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2012
2013#ifdef CONFIG_SCHED_DEBUG
2014extern unsigned int sysctl_sched_migration_cost;
2015extern unsigned int sysctl_sched_nr_migrate;
2016extern unsigned int sysctl_sched_time_avg;
2017extern unsigned int sysctl_timer_migration;
2018extern unsigned int sysctl_sched_shares_window;
2019
2020int sched_proc_update_handler(struct ctl_table *table, int write,
2021                void __user *buffer, size_t *length,
2022                loff_t *ppos);
2023#endif
2024#ifdef CONFIG_SCHED_DEBUG
2025static inline unsigned int get_sysctl_timer_migration(void)
2026{
2027        return sysctl_timer_migration;
2028}
2029#else
2030static inline unsigned int get_sysctl_timer_migration(void)
2031{
2032        return 1;
2033}
2034#endif
2035extern unsigned int sysctl_sched_rt_period;
2036extern int sysctl_sched_rt_runtime;
2037
2038int sched_rt_handler(struct ctl_table *table, int write,
2039                void __user *buffer, size_t *lenp,
2040                loff_t *ppos);
2041
2042#ifdef CONFIG_SCHED_AUTOGROUP
2043extern unsigned int sysctl_sched_autogroup_enabled;
2044
2045extern void sched_autogroup_create_attach(struct task_struct *p);
2046extern void sched_autogroup_detach(struct task_struct *p);
2047extern void sched_autogroup_fork(struct signal_struct *sig);
2048extern void sched_autogroup_exit(struct signal_struct *sig);
2049#ifdef CONFIG_PROC_FS
2050extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2051extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2052#endif
2053#else
2054static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2055static inline void sched_autogroup_detach(struct task_struct *p) { }
2056static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2057static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2058#endif
2059
2060#ifdef CONFIG_CFS_BANDWIDTH
2061extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2062#endif
2063
2064#ifdef CONFIG_RT_MUTEXES
2065extern int rt_mutex_getprio(struct task_struct *p);
2066extern void rt_mutex_setprio(struct task_struct *p, int prio);
2067extern void rt_mutex_adjust_pi(struct task_struct *p);
2068static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2069{
2070        return tsk->pi_blocked_on != NULL;
2071}
2072#else
2073static inline int rt_mutex_getprio(struct task_struct *p)
2074{
2075        return p->normal_prio;
2076}
2077# define rt_mutex_adjust_pi(p)          do { } while (0)
2078static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2079{
2080        return false;
2081}
2082#endif
2083
2084extern bool yield_to(struct task_struct *p, bool preempt);
2085extern void set_user_nice(struct task_struct *p, long nice);
2086extern int task_prio(const struct task_struct *p);
2087extern int task_nice(const struct task_struct *p);
2088extern int can_nice(const struct task_struct *p, const int nice);
2089extern int task_curr(const struct task_struct *p);
2090extern int idle_cpu(int cpu);
2091extern int sched_setscheduler(struct task_struct *, int,
2092                              const struct sched_param *);
2093extern int sched_setscheduler_nocheck(struct task_struct *, int,
2094                                      const struct sched_param *);
2095extern struct task_struct *idle_task(int cpu);
2096/**
2097 * is_idle_task - is the specified task an idle task?
2098 * @p: the task in question.
2099 */
2100static inline bool is_idle_task(const struct task_struct *p)
2101{
2102        return p->pid == 0;
2103}
2104extern struct task_struct *curr_task(int cpu);
2105extern void set_curr_task(int cpu, struct task_struct *p);
2106
2107void yield(void);
2108
2109/*
2110 * The default (Linux) execution domain.
2111 */
2112extern struct exec_domain       default_exec_domain;
2113
2114union thread_union {
2115        struct thread_info thread_info;
2116        unsigned long stack[THREAD_SIZE/sizeof(long)];
2117};
2118
2119#ifndef __HAVE_ARCH_KSTACK_END
2120static inline int kstack_end(void *addr)
2121{
2122        /* Reliable end of stack detection:
2123         * Some APM bios versions misalign the stack
2124         */
2125        return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2126}
2127#endif
2128
2129extern union thread_union init_thread_union;
2130extern struct task_struct init_task;
2131
2132extern struct   mm_struct init_mm;
2133
2134extern struct pid_namespace init_pid_ns;
2135
2136/*
2137 * find a task by one of its numerical ids
2138 *
2139 * find_task_by_pid_ns():
2140 *      finds a task by its pid in the specified namespace
2141 * find_task_by_vpid():
2142 *      finds a task by its virtual pid
2143 *
2144 * see also find_vpid() etc in include/linux/pid.h
2145 */
2146
2147extern struct task_struct *find_task_by_vpid(pid_t nr);
2148extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2149                struct pid_namespace *ns);
2150
2151extern void __set_special_pids(struct pid *pid);
2152
2153/* per-UID process charging. */
2154extern struct user_struct * alloc_uid(kuid_t);
2155static inline struct user_struct *get_uid(struct user_struct *u)
2156{
2157        atomic_inc(&u->__count);
2158        return u;
2159}
2160extern void free_uid(struct user_struct *);
2161
2162#include <asm/current.h>
2163
2164extern void xtime_update(unsigned long ticks);
2165
2166extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2167extern int wake_up_process(struct task_struct *tsk);
2168extern void wake_up_new_task(struct task_struct *tsk);
2169#ifdef CONFIG_SMP
2170 extern void kick_process(struct task_struct *tsk);
2171#else
2172 static inline void kick_process(struct task_struct *tsk) { }
2173#endif
2174extern void sched_fork(struct task_struct *p);
2175extern void sched_dead(struct task_struct *p);
2176
2177extern void proc_caches_init(void);
2178extern void flush_signals(struct task_struct *);
2179extern void __flush_signals(struct task_struct *);
2180extern void ignore_signals(struct task_struct *);
2181extern void flush_signal_handlers(struct task_struct *, int force_default);
2182extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2183
2184static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2185{
2186        unsigned long flags;
2187        int ret;
2188
2189        spin_lock_irqsave(&tsk->sighand->siglock, flags);
2190        ret = dequeue_signal(tsk, mask, info);
2191        spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2192
2193        return ret;
2194}
2195
2196extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2197                              sigset_t *mask);
2198extern void unblock_all_signals(void);
2199extern void release_task(struct task_struct * p);
2200extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2201extern int force_sigsegv(int, struct task_struct *);
2202extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2203extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2204extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2205extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2206                                const struct cred *, u32);
2207extern int kill_pgrp(struct pid *pid, int sig, int priv);
2208extern int kill_pid(struct pid *pid, int sig, int priv);
2209extern int kill_proc_info(int, struct siginfo *, pid_t);
2210extern __must_check bool do_notify_parent(struct task_struct *, int);
2211extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2212extern void force_sig(int, struct task_struct *);
2213extern int send_sig(int, struct task_struct *, int);
2214extern int zap_other_threads(struct task_struct *p);
2215extern struct sigqueue *sigqueue_alloc(void);
2216extern void sigqueue_free(struct sigqueue *);
2217extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2218extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2219extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2220
2221static inline void restore_saved_sigmask(void)
2222{
2223        if (test_and_clear_restore_sigmask())
2224                __set_current_blocked(&current->saved_sigmask);
2225}
2226
2227static inline sigset_t *sigmask_to_save(void)
2228{
2229        sigset_t *res = &current->blocked;
2230        if (unlikely(test_restore_sigmask()))
2231                res = &current->saved_sigmask;
2232        return res;
2233}
2234
2235static inline int kill_cad_pid(int sig, int priv)
2236{
2237        return kill_pid(cad_pid, sig, priv);
2238}
2239
2240/* These can be the second arg to send_sig_info/send_group_sig_info.  */
2241#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2242#define SEND_SIG_PRIV   ((struct siginfo *) 1)
2243#define SEND_SIG_FORCED ((struct siginfo *) 2)
2244
2245/*
2246 * True if we are on the alternate signal stack.
2247 */
2248static inline int on_sig_stack(unsigned long sp)
2249{
2250#ifdef CONFIG_STACK_GROWSUP
2251        return sp >= current->sas_ss_sp &&
2252                sp - current->sas_ss_sp < current->sas_ss_size;
2253#else
2254        return sp > current->sas_ss_sp &&
2255                sp - current->sas_ss_sp <= current->sas_ss_size;
2256#endif
2257}
2258
2259static inline int sas_ss_flags(unsigned long sp)
2260{
2261        return (current->sas_ss_size == 0 ? SS_DISABLE
2262                : on_sig_stack(sp) ? SS_ONSTACK : 0);
2263}
2264
2265/*
2266 * Routines for handling mm_structs
2267 */
2268extern struct mm_struct * mm_alloc(void);
2269
2270/* mmdrop drops the mm and the page tables */
2271extern void __mmdrop(struct mm_struct *);
2272static inline void mmdrop(struct mm_struct * mm)
2273{
2274        if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2275                __mmdrop(mm);
2276}
2277
2278/* mmput gets rid of the mappings and all user-space */
2279extern void mmput(struct mm_struct *);
2280/* Grab a reference to a task's mm, if it is not already going away */
2281extern struct mm_struct *get_task_mm(struct task_struct *task);
2282/*
2283 * Grab a reference to a task's mm, if it is not already going away
2284 * and ptrace_may_access with the mode parameter passed to it
2285 * succeeds.
2286 */
2287extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2288/* Remove the current tasks stale references to the old mm_struct */
2289extern void mm_release(struct task_struct *, struct mm_struct *);
2290/* Allocate a new mm structure and copy contents from tsk->mm */
2291extern struct mm_struct *dup_mm(struct task_struct *tsk);
2292
2293extern int copy_thread(unsigned long, unsigned long, unsigned long,
2294                        struct task_struct *, struct pt_regs *);
2295extern void flush_thread(void);
2296extern void exit_thread(void);
2297
2298extern void exit_files(struct task_struct *);
2299extern void __cleanup_sighand(struct sighand_struct *);
2300
2301extern void exit_itimers(struct signal_struct *);
2302extern void flush_itimer_signals(void);
2303
2304extern void do_group_exit(int);
2305
2306extern void daemonize(const char *, ...);
2307extern int allow_signal(int);
2308extern int disallow_signal(int);
2309
2310extern int do_execve(const char *,
2311                     const char __user * const __user *,
2312                     const char __user * const __user *, struct pt_regs *);
2313extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2314struct task_struct *fork_idle(int);
2315
2316extern void set_task_comm(struct task_struct *tsk, char *from);
2317extern char *get_task_comm(char *to, struct task_struct *tsk);
2318
2319#ifdef CONFIG_SMP
2320void scheduler_ipi(void);
2321extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2322#else
2323static inline void scheduler_ipi(void) { }
2324static inline unsigned long wait_task_inactive(struct task_struct *p,
2325                                               long match_state)
2326{
2327        return 1;
2328}
2329#endif
2330
2331#define next_task(p) \
2332        list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2333
2334#define for_each_process(p) \
2335        for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2336
2337extern bool current_is_single_threaded(void);
2338
2339/*
2340 * Careful: do_each_thread/while_each_thread is a double loop so
2341 *          'break' will not work as expected - use goto instead.
2342 */
2343#define do_each_thread(g, t) \
2344        for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2345
2346#define while_each_thread(g, t) \
2347        while ((t = next_thread(t)) != g)
2348
2349static inline int get_nr_threads(struct task_struct *tsk)
2350{
2351        return tsk->signal->nr_threads;
2352}
2353
2354static inline bool thread_group_leader(struct task_struct *p)
2355{
2356        return p->exit_signal >= 0;
2357}
2358
2359/* Do to the insanities of de_thread it is possible for a process
2360 * to have the pid of the thread group leader without actually being
2361 * the thread group leader.  For iteration through the pids in proc
2362 * all we care about is that we have a task with the appropriate
2363 * pid, we don't actually care if we have the right task.
2364 */
2365static inline int has_group_leader_pid(struct task_struct *p)
2366{
2367        return p->pid == p->tgid;
2368}
2369
2370static inline
2371int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2372{
2373        return p1->tgid == p2->tgid;
2374}
2375
2376static inline struct task_struct *next_thread(const struct task_struct *p)
2377{
2378        return list_entry_rcu(p->thread_group.next,
2379                              struct task_struct, thread_group);
2380}
2381
2382static inline int thread_group_empty(struct task_struct *p)
2383{
2384        return list_empty(&p->thread_group);
2385}
2386
2387#define delay_group_leader(p) \
2388                (thread_group_leader(p) && !thread_group_empty(p))
2389
2390/*
2391 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2392 * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2393 * pins the final release of task.io_context.  Also protects ->cpuset and
2394 * ->cgroup.subsys[]. And ->vfork_done.
2395 *
2396 * Nests both inside and outside of read_lock(&tasklist_lock).
2397 * It must not be nested with write_lock_irq(&tasklist_lock),
2398 * neither inside nor outside.
2399 */
2400static inline void task_lock(struct task_struct *p)
2401{
2402        spin_lock(&p->alloc_lock);
2403}
2404
2405static inline void task_unlock(struct task_struct *p)
2406{
2407        spin_unlock(&p->alloc_lock);
2408}
2409
2410extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2411                                                        unsigned long *flags);
2412
2413static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2414                                                       unsigned long *flags)
2415{
2416        struct sighand_struct *ret;
2417
2418        ret = __lock_task_sighand(tsk, flags);
2419        (void)__cond_lock(&tsk->sighand->siglock, ret);
2420        return ret;
2421}
2422
2423static inline void unlock_task_sighand(struct task_struct *tsk,
2424                                                unsigned long *flags)
2425{
2426        spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2427}
2428
2429#ifdef CONFIG_CGROUPS
2430static inline void threadgroup_change_begin(struct task_struct *tsk)
2431{
2432        down_read(&tsk->signal->group_rwsem);
2433}
2434static inline void threadgroup_change_end(struct task_struct *tsk)
2435{
2436        up_read(&tsk->signal->group_rwsem);
2437}
2438
2439/**
2440 * threadgroup_lock - lock threadgroup
2441 * @tsk: member task of the threadgroup to lock
2442 *
2443 * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2444 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2445 * perform exec.  This is useful for cases where the threadgroup needs to
2446 * stay stable across blockable operations.
2447 *
2448 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2449 * synchronization.  While held, no new task will be added to threadgroup
2450 * and no existing live task will have its PF_EXITING set.
2451 *
2452 * During exec, a task goes and puts its thread group through unusual
2453 * changes.  After de-threading, exclusive access is assumed to resources
2454 * which are usually shared by tasks in the same group - e.g. sighand may
2455 * be replaced with a new one.  Also, the exec'ing task takes over group
2456 * leader role including its pid.  Exclude these changes while locked by
2457 * grabbing cred_guard_mutex which is used to synchronize exec path.
2458 */
2459static inline void threadgroup_lock(struct task_struct *tsk)
2460{
2461        /*
2462         * exec uses exit for de-threading nesting group_rwsem inside
2463         * cred_guard_mutex. Grab cred_guard_mutex first.
2464         */
2465        mutex_lock(&tsk->signal->cred_guard_mutex);
2466        down_write(&tsk->signal->group_rwsem);
2467}
2468
2469/**
2470 * threadgroup_unlock - unlock threadgroup
2471 * @tsk: member task of the threadgroup to unlock
2472 *
2473 * Reverse threadgroup_lock().
2474 */
2475static inline void threadgroup_unlock(struct task_struct *tsk)
2476{
2477        up_write(&tsk->signal->group_rwsem);
2478        mutex_unlock(&tsk->signal->cred_guard_mutex);
2479}
2480#else
2481static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2482static inline void threadgroup_change_end(struct task_struct *tsk) {}
2483static inline void threadgroup_lock(struct task_struct *tsk) {}
2484static inline void threadgroup_unlock(struct task_struct *tsk) {}
2485#endif
2486
2487#ifndef __HAVE_THREAD_FUNCTIONS
2488
2489#define task_thread_info(task)  ((struct thread_info *)(task)->stack)
2490#define task_stack_page(task)   ((task)->stack)
2491
2492static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2493{
2494        *task_thread_info(p) = *task_thread_info(org);
2495        task_thread_info(p)->task = p;
2496}
2497
2498static inline unsigned long *end_of_stack(struct task_struct *p)
2499{
2500        return (unsigned long *)(task_thread_info(p) + 1);
2501}
2502
2503#endif
2504
2505static inline int object_is_on_stack(void *obj)
2506{
2507        void *stack = task_stack_page(current);
2508
2509        return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2510}
2511
2512extern void thread_info_cache_init(void);
2513
2514#ifdef CONFIG_DEBUG_STACK_USAGE
2515static inline unsigned long stack_not_used(struct task_struct *p)
2516{
2517        unsigned long *n = end_of_stack(p);
2518
2519        do {    /* Skip over canary */
2520                n++;
2521        } while (!*n);
2522
2523        return (unsigned long)n - (unsigned long)end_of_stack(p);
2524}
2525#endif
2526
2527/* set thread flags in other task's structures
2528 * - see asm/thread_info.h for TIF_xxxx flags available
2529 */
2530static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2531{
2532        set_ti_thread_flag(task_thread_info(tsk), flag);
2533}
2534
2535static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2536{
2537        clear_ti_thread_flag(task_thread_info(tsk), flag);
2538}
2539
2540static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2541{
2542        return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2543}
2544
2545static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2546{
2547        return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2548}
2549
2550static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2551{
2552        return test_ti_thread_flag(task_thread_info(tsk), flag);
2553}
2554
2555static inline void set_tsk_need_resched(struct task_struct *tsk)
2556{
2557        set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2558}
2559
2560static inline void clear_tsk_need_resched(struct task_struct *tsk)
2561{
2562        clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2563}
2564
2565static inline int test_tsk_need_resched(struct task_struct *tsk)
2566{
2567        return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2568}
2569
2570static inline int restart_syscall(void)
2571{
2572        set_tsk_thread_flag(current, TIF_SIGPENDING);
2573        return -ERESTARTNOINTR;
2574}
2575
2576static inline int signal_pending(struct task_struct *p)
2577{
2578        return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2579}
2580
2581static inline int __fatal_signal_pending(struct task_struct *p)
2582{
2583        return unlikely(sigismember(&p->pending.signal, SIGKILL));
2584}
2585
2586static inline int fatal_signal_pending(struct task_struct *p)
2587{
2588        return signal_pending(p) && __fatal_signal_pending(p);
2589}
2590
2591static inline int signal_pending_state(long state, struct task_struct *p)
2592{
2593        if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2594                return 0;
2595        if (!signal_pending(p))
2596                return 0;
2597
2598        return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2599}
2600
2601static inline int need_resched(void)
2602{
2603        return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2604}
2605
2606/*
2607 * cond_resched() and cond_resched_lock(): latency reduction via
2608 * explicit rescheduling in places that are safe. The return
2609 * value indicates whether a reschedule was done in fact.
2610 * cond_resched_lock() will drop the spinlock before scheduling,
2611 * cond_resched_softirq() will enable bhs before scheduling.
2612 */
2613extern int _cond_resched(void);
2614
2615#define cond_resched() ({                       \
2616        __might_sleep(__FILE__, __LINE__, 0);   \
2617        _cond_resched();                        \
2618})
2619
2620extern int __cond_resched_lock(spinlock_t *lock);
2621
2622#ifdef CONFIG_PREEMPT_COUNT
2623#define PREEMPT_LOCK_OFFSET     PREEMPT_OFFSET
2624#else
2625#define PREEMPT_LOCK_OFFSET     0
2626#endif
2627
2628#define cond_resched_lock(lock) ({                              \
2629        __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2630        __cond_resched_lock(lock);                              \
2631})
2632
2633extern int __cond_resched_softirq(void);
2634
2635#define cond_resched_softirq() ({                                       \
2636        __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);      \
2637        __cond_resched_softirq();                                       \
2638})
2639
2640/*
2641 * Does a critical section need to be broken due to another
2642 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2643 * but a general need for low latency)
2644 */
2645static inline int spin_needbreak(spinlock_t *lock)
2646{
2647#ifdef CONFIG_PREEMPT
2648        return spin_is_contended(lock);
2649#else
2650        return 0;
2651#endif
2652}
2653
2654/*
2655 * Thread group CPU time accounting.
2656 */
2657void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2658void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2659
2660static inline void thread_group_cputime_init(struct signal_struct *sig)
2661{
2662        raw_spin_lock_init(&sig->cputimer.lock);
2663}
2664
2665/*
2666 * Reevaluate whether the task has signals pending delivery.
2667 * Wake the task if so.
2668 * This is required every time the blocked sigset_t changes.
2669 * callers must hold sighand->siglock.
2670 */
2671extern void recalc_sigpending_and_wake(struct task_struct *t);
2672extern void recalc_sigpending(void);
2673
2674extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2675
2676/*
2677 * Wrappers for p->thread_info->cpu access. No-op on UP.
2678 */
2679#ifdef CONFIG_SMP
2680
2681static inline unsigned int task_cpu(const struct task_struct *p)
2682{
2683        return task_thread_info(p)->cpu;
2684}
2685
2686extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2687
2688#else
2689
2690static inline unsigned int task_cpu(const struct task_struct *p)
2691{
2692        return 0;
2693}
2694
2695static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2696{
2697}
2698
2699#endif /* CONFIG_SMP */
2700
2701extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2702extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2703
2704extern void normalize_rt_tasks(void);
2705
2706#ifdef CONFIG_CGROUP_SCHED
2707
2708extern struct task_group root_task_group;
2709
2710extern struct task_group *sched_create_group(struct task_group *parent);
2711extern void sched_destroy_group(struct task_group *tg);
2712extern void sched_move_task(struct task_struct *tsk);
2713#ifdef CONFIG_FAIR_GROUP_SCHED
2714extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2715extern unsigned long sched_group_shares(struct task_group *tg);
2716#endif
2717#ifdef CONFIG_RT_GROUP_SCHED
2718extern int sched_group_set_rt_runtime(struct task_group *tg,
2719                                      long rt_runtime_us);
2720extern long sched_group_rt_runtime(struct task_group *tg);
2721extern int sched_group_set_rt_period(struct task_group *tg,
2722                                      long rt_period_us);
2723extern long sched_group_rt_period(struct task_group *tg);
2724extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2725#endif
2726#endif
2727
2728extern int task_can_switch_user(struct user_struct *up,
2729                                        struct task_struct *tsk);
2730
2731#ifdef CONFIG_TASK_XACCT
2732static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2733{
2734        tsk->ioac.rchar += amt;
2735}
2736
2737static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2738{
2739        tsk->ioac.wchar += amt;
2740}
2741
2742static inline void inc_syscr(struct task_struct *tsk)
2743{
2744        tsk->ioac.syscr++;
2745}
2746
2747static inline void inc_syscw(struct task_struct *tsk)
2748{
2749        tsk->ioac.syscw++;
2750}
2751#else
2752static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2753{
2754}
2755
2756static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2757{
2758}
2759
2760static inline void inc_syscr(struct task_struct *tsk)
2761{
2762}
2763
2764static inline void inc_syscw(struct task_struct *tsk)
2765{
2766}
2767#endif
2768
2769#ifndef TASK_SIZE_OF
2770#define TASK_SIZE_OF(tsk)       TASK_SIZE
2771#endif
2772
2773#ifdef CONFIG_MM_OWNER
2774extern void mm_update_next_owner(struct mm_struct *mm);
2775extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2776#else
2777static inline void mm_update_next_owner(struct mm_struct *mm)
2778{
2779}
2780
2781static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2782{
2783}
2784#endif /* CONFIG_MM_OWNER */
2785
2786static inline unsigned long task_rlimit(const struct task_struct *tsk,
2787                unsigned int limit)
2788{
2789        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2790}
2791
2792static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2793                unsigned int limit)
2794{
2795        return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2796}
2797
2798static inline unsigned long rlimit(unsigned int limit)
2799{
2800        return task_rlimit(current, limit);
2801}
2802
2803static inline unsigned long rlimit_max(unsigned int limit)
2804{
2805        return task_rlimit_max(current, limit);
2806}
2807
2808#endif /* __KERNEL__ */
2809
2810#endif
2811
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.